Successive Screening¶
Filtering procedure — instantiates Progressive Narrowing
Makes an unmanageably large pool tractable by applying a sequence of filters — cheapest and most discriminating first, deeper and costlier later — so each reviewable stage hands the next a set it can actually afford to examine.
Successive Screening applies a series of filters one after another to a pool too large to evaluate at full depth, ordered so that the cheapest, most discriminating tests run first and remove the clear mismatches, leaving progressively smaller sets for progressively deeper review. Its defining concern is tractability: it exists because the pool cannot be examined all at once, so its job is to shed volume at each stage cheaply and reviewably until what remains fits the attention available, with each cut owned by a visible gate. It is narrowing as load management, not as final judgment.
Example¶
A space telescope monitoring 200,000 stars produces millions of brightness dips that could be transiting planets. No team can inspect each by hand, so the survey runs a screening sequence. A first automated filter demands a periodic, transit-shaped dip above a signal threshold, cutting millions of noise events to thousands of "threshold-crossing events." A second screen removes the obvious false positives — eclipsing binary stars, whose deep V-shaped dips mimic planets — and instrumental systematics. A third stage applies statistical validation to the survivors, estimating the odds each is a genuine planet rather than a background blend. Only the few hundred that clear every screen reach expensive follow-up: radial-velocity observations on large ground telescopes to confirm and weigh the planet.
Each screen is cheap relative to the next and hands forward a set the next stage can afford; each cut is a recorded gate, so a later reanalysis can ask whether the first threshold was throwing away small real planets. The value is turning an impossible inspection into a sequence of affordable ones.
How it works¶
- Order screens by cost and discrimination. The first screen is the cheapest test that removes the most clear non-matches; depth and cost rise down the sequence.
- Size each cut to the next stage's capacity. Each screen forwards a set the following, costlier stage can actually process.
- Gate every cut. Each stage's advance/reject is an owned, recorded decision rather than an automatic disappearance, so the reduction stays inspectable.
- Reserve depth for survivors. Expensive evaluation is spent only on what has already survived the cheap screens.
Tuning parameters¶
- Screen ordering — which filter runs first. Front-loading the most discriminating cheap screen sheds the most volume early; a poor order pays deep-stage cost on items a later-but-cheaper screen would have removed.
- Per-screen strictness — how aggressively each filter cuts. Strict screens protect downstream capacity but raise the false-negative rate; loose screens preserve recall but leave too many survivors.
- Number of stages — how many screens in the sequence. More stages give a smoother reduction and more chances to catch value but add handoff overhead.
- Attention-budget allocation — how much review capacity each stage receives. Concentrating budget late deepens final review but starves early discrimination.
When it helps, and when it misleads¶
Its strength is that it makes an otherwise impossible evaluation affordable, spending scarce deep-review attention only on candidates that have already survived cheap discrimination, while keeping each cut visible.
Its failure mode comes from the early screens doing load-bearing work on thin evidence. A cheap first filter carries a real false-negative rate — a genuine match discarded as noise never reaches the discriminating stages, a Type II error[n1] baked into the pipeline — and the pressure to make the pool tractable tempts a team to tighten a screen until the survivor count hits a target, optimizing the pool's size rather than its quality. The classic misuse is treating how much was cut as the measure of success. The guarding discipline is to re-test a sample of rejects against the final criterion, calibrate early screens against known-good exemplars, and judge the sequence by end-of-pipeline yield rather than by how much it removed.
How it implements the components¶
option_set— it operates on an explicit, very large pool of candidates; the whole point is that the set exceeds what can be reviewed directly.filter_sequence— the ordered series of filters, cheapest and most discriminating first and deeper later, is its core structure.decision_gate— each screen's advance/reject is an owned, recorded gate rather than an automatic disappearance.capacity_or_attention_budget— each cut is sized to the review capacity of the next stage, so scarce deep-evaluation attention is never overrun.
It makes a pool tractable but attaches no deep evidence_requirement such as a structured interview and fires no finalist_commitment_rule — those are hiring_shortlist_process — and writes no per-candidate elimination_rationale_record, which is candidate_disposition_log; the screens produce cuts, the log records them.
Related¶
- Instantiates: Progressive Narrowing — cheap-first, capacity-sized filtering of an oversized pool into a reviewable survivor set.
- Sibling mechanisms: Hiring Shortlist Process · Procurement Shortlisting · Diagnostic Narrowing Protocol · Research Hypothesis Elimination · Design Downselection Review · Legal Issue Narrowing · Candidate Disposition Log · Weighted Scoring Matrix · Funnel Process
Editorial Notes¶
Form Classification¶
Form family: Decision, Gate & Allocation
Rationale: Successive Screening is defined in the frozen evidence as: Makes an unmanageably large pool tractable by applying a sequence of filters — cheapest and most discriminating first, deeper and costlier later — so each reviewable stage hands the next a set it can actually afford to examine. Its operative deployed or enacted form is therefore Decision, Gate & Allocation.
Nearest alternative: Protocol, Workflow & Routine — Protocol, Workflow & Routine can support this mechanism, but the evidence centers the concrete operation described above rather than the alternative family's defining operation.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Operations Research
Origin pattern: Convergent development
Present-day reach: Universal
Rationale: Allocating small initial tests to many candidates and progressively concentrating resources on survivors is sequential screening and elimination. Successive-halving research formalizes staged resource allocation and pruning; experimental screening designs provide an independent lineage.
Related originating lineages:
- Computer Science & Software Engineering — computer_science contributes computer science and software-engineering practice to this mechanism's defining operation—Makes an unmanageably large pool tractable by applying a sequence of filters — cheapest and most discriminating first, deeper and costlier later — so each reviewable stage hands the next a set it can actually afford to examine—without displacing the selected primary historical lineage.
- Data Science & Analytics — data_science contributes data science, analytics, and operational monitoring to this mechanism's defining operation—Makes an unmanageably large pool tractable by applying a sequence of filters — cheapest and most discriminating first, deeper and costlier later — so each reviewable stage hands the next a set it can actually afford to examine—without displacing the selected primary historical lineage.
- Engineering & Design — engineering_design contributes engineering design, reliability, and systems-safety practice to this mechanism's defining operation—Makes an unmanageably large pool tractable by applying a sequence of filters — cheapest and most discriminating first, deeper and costlier later — so each reviewable stage hands the next a set it can actually afford to examine—without displacing the selected primary historical lineage.
- Medicine & Healthcare — Diagnostic screening uses sequential tests of increasing specificity.
- Organizational & Management Science — Hiring and review funnels allocate scarce attention in stages.
- Statistics & Experimental Design — Statistics, experimental design, and measurement theory supplies a parallel or contributing lineage for the mechanism's defining operation: makes an unmanageably large pool tractable by applying a sequence of filters — cheapest and most discriminating first, deeper and costlier later — so each reviewable stage hands the….
- Systems Thinking & Cybernetics — Systems thinking, feedback control, and cybernetics supplies a parallel or contributing lineage for the mechanism's defining operation: makes an unmanageably large pool tractable by applying a sequence of filters — cheapest and most discriminating first, deeper and costlier later — so each reviewable stage hands the….
Review resolution: The blind reviewers disagree on primary lineage (operations_research versus organizational_management). Authoritative or primary research supports operations_research as the best historical origin: Allocating small initial tests to many candidates and progressively concentrating resources on survivors is sequential screening and elimination. Successive-halving research formalizes staged resource allocation and pruning; experimental screening designs provide an independent lineage. The cited Successive Halving for Hyperparameter Optimization; NIH/PubMed Central, Successive Screening in Experimental Design directly supports the mechanism's defining operation. All independently supported contributing domains are retained without an arbitrary cap. origin_mode=convergent records lineage, while domain_reach=universal records later applicability separately from provenance.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
- Successive Halving for Hyperparameter Optimization
- NIH/PubMed Central, Successive Screening in Experimental Design
Notes¶
Successive Screening and the Funnel Process are close cousins, and the difference is one of emphasis. The funnel engineers per-stage cost and pass-rate economics so no stage is flooded and no expensive capacity sits idle. Successive Screening centers on shedding volume cheaply and reviewably to make an oversized pool tractable — throughput balance is welcome but secondary to keeping each cut discriminating and inspectable.
[n1] Type II error — in statistics, a false negative: failing to detect a real effect (here, a genuine match) that is actually present. Cheap high-throughput screens trade a higher Type II rate for speed, which is exactly why sampling and re-testing the rejected pool is the standard check on a screening sequence. ↩